Forecasting of Photovoltaic Power Generation: Techniques and Key Factors

The aim of this paper is first, to provide a detailed overview on conventional and modern techniques of solar photovoltaic (PV) forecasting and then, to determine the predominating factors that affect the generation of PV power. To mitigate the depletion of fossil fuel and reduce greenhouse gas emissions, renewable energy is being deployed all over the world. Amongst various options, PV is one of the leading renewable generation sources. Due to variability, uncertainty and dependence on weather condition, PV power forecasting is treated as a challenging task. To this end, several PV forecasting techniques have been developed. Before the widespread application of machine learning tools, conventional methods viz. statistical, autoregressive moving average, regression etc. were used to forecast PV power. However, with recent advancement of machine learning approaches and access to big data, Artificial Neural network, Support Vector machine and hybrid techniques have become prevalent. To better understand the insights of PV prediction, this paper explores the working principle and applicability of various PV forecasting methods. In addition, the key factors affecting the PV generation are estimated using practical data.